Quantification of Interfraction Gastric Motion by Daily Magnetic Resonance Imaging and Implications for Radiation Treatment Planning

Purpose/Objective(s) Accurate radiation therapy targeting for stomach cancer is challenging given daily variations in the stomach's size and position. We evaluated daily MR-based imaging to quantify changes in stomach volume and characterized the impact of inter-fractional variation on dosimetry for gastric cancer radiation plans. Materials/Methods We assessed inter-fraction changes in the volume and position of the stomach via MR-based imaging for 10 patients who previously underwent abdominal stereotactic body radiation therapy. All patients were simulated and treated over 5 fractions and were instructed to fast for 3 hours before each session. The stomach and regional organs-at-risk (OARs) were contoured on each fraction's MR scan according to standard RTOG guidelines. Inter-fraction variation in the size and position of the stomach was quantified by calculating the percentage volume change and dice similarity coefficient (DSC) as compared to the simulation scan. To determine the potential impact of inter-fraction stomach motion on treatment planning, we generated mock plans to recapitulate neoadjuvant or definitive stomach cancer treatments. For the purpose of this study, the stomach contour was symmetrically expanded by 1.7 cm to generate a planning treatment volume (PTV) and a prescription dose of 45 Gy in 1.8 Gy fractions was applied. Intensity-modulated radiation therapy plans were then generated to meet the following dosimetric goals: PTV V95% of 95%, liver V30 less than 30%, and bilateral kidney V20 less than 20%. Finally, the impact of gastric variation on these dose metrics was assessed. Results Despite standard daily preparation and set up procedures, we observed significant variability in the size and location of the stomach between fractions. The mean absolute daily change in gastric volume compared to the planning MRI was 19.1% (total range: -42.0% to 58.4%). The average stomach DSC was 0.69 (range: 0.40 to 0.86). Given this degree of variability, mock gastric cancer treatment plans displayed suboptimal dose metrics. The average PTV V45 was 81.5% (range: 58.8% to 99.2%. 95% CI: 78.8% to 84.2%). Average liver V30 values were 12.7% (range: 3.5% to 20.1%. 95% CI: 11.7% to 13.9%) and bilateral kidney V20 values were 12.3% (range: 0.5% to 31.3%. 95% CI: 9.2% to 15.3%). Target metrics were highly correlated with the percent change in stomach volume—a 10% increase in stomach volume was associated with a 2.4% decrease in the V45 (P Conclusion We observed significant inter-fraction variability in the dimensions of the stomach using daily MRI-based imaging, leading to dosimetrically inferior mock gastric cancer treatment plans. Our study suggests that larger PTV margins may be necessary for conventional stomach plans, which would come at the expense of greater OAR exposure. Alternatively, MRI-based image guidance and daily adaptive treatments may improve the therapeutic window of radiation therapy for gastric tumors.